Azure AI Content Safety client library for Python - version 1.0.0

Azure AI Content Safety detects harmful user-generated and AI-generated content in applications and services. Content Safety includes text and image APIs that allow you to detect material that is harmful:

  • Text Analysis API: Scans text for sexual content, violence, hate, and self-harm with multi-severity levels.
  • Image Analysis API: Scans images for sexual content, violence, hate, and self-harm with multi-severity levels.
  • Text Blocklist Management APIs: The default AI classifiers are sufficient for most content safety needs; however, you might need to screen for terms that are specific to your use case. You can create blocklists of terms to use with the Text API.

Documentation

Various documentation is available to help you get started

Getting started

Prerequisites

Install the package

pip install azure-ai-contentsafety

Authenticate the client

Get the endpoint

You can find the endpoint for your Azure AI Content Safety service resource using the Azure Portal or Azure CLI:

# Get the endpoint for the Azure AI Content Safety service resource
az cognitiveservices account show --name "resource-name" --resource-group "resource-group-name" --query "properties.endpoint"

Create a ContentSafetyClient/BlocklistClient with API key

To use an API key as the credential parameter.

  • Step 1: Get the API key. The API key can be found in the Azure Portal or by running the following Azure CLI command:

    az cognitiveservices account keys list --name "<resource-name>" --resource-group "<resource-group-name>"
    
  • Step 2: Pass the key as a string into an instance of AzureKeyCredential.

    from azure.core.credentials import AzureKeyCredential
    from azure.ai.contentsafety import ContentSafetyClient, BlocklistClient
    
    endpoint = "https://<my-custom-subdomain>.cognitiveservices.azure.com/"
    credential = AzureKeyCredential("<api_key>")
    content_safety_client = ContentSafetyClient(endpoint, credential)
    blocklist_client = BlocklistClient(endpoint, credential)
    

Create a ContentSafetyClient/BlocklistClient with Microsoft Entra ID token credential

  • Step 1: Enable Microsoft Entra ID for your resource. Please refer to this document Authenticate with Microsoft Entra ID for the steps to enable Microsoft Entra ID for your resource.

    The main steps are:

    • Create resource with a custom subdomain.
    • Create Service Principal and assign Cognitive Services User role to it.
  • Step 2: Set the values of the client ID, tenant ID, and client secret of the Microsoft Entra application as environment variables: AZURE_CLIENT_ID, AZURE_TENANT_ID, AZURE_CLIENT_SECRET.

    DefaultAzureCredential will use the values from these environment variables.

    from azure.identity import DefaultAzureCredential
    from azure.ai.contentsafety import ContentSafetyClient, BlocklistClient
    
    endpoint = "https://<my-custom-subdomain>.cognitiveservices.azure.com/"
    credential = DefaultAzureCredential()
    content_safety_client = ContentSafetyClient(endpoint, credential)
    blocklist_client = BlocklistClient(endpoint, credential)
    

Key concepts

Available features

There are different types of analysis available from this service. The following table describes the currently available APIs.

Feature Description
Text Analysis API Scans text for sexual content, violence, hate, and self-harm with multi-severity levels.
Image Analysis API Scans images for sexual content, violence, hate, and self-harm with multi-severity levels.
Text Blocklist Management APIs The default AI classifiers are sufficient for most content safety needs. However, you might need to screen for terms that are specific to your use case. You can create blocklists of terms to use with the Text API.

Harm categories

Content Safety recognizes four distinct categories of objectionable content.

Category Description
Hate Hate and fairness-related harms refer to any content that attacks or uses pejorative or discriminatory language with reference to a person or identity group based on certain differentiating attributes of these groups including but not limited to race, ethnicity, nationality, gender identity and expression, sexual orientation, religion, immigration status, ability status, personal appearance, and body size.
Sexual Sexual describes language related to anatomical organs and genitals, romantic relationships, acts portrayed in erotic or affectionate terms, pregnancy, physical sexual acts, including those portrayed as an assault or a forced sexual violent act against one's will, prostitution, pornography, and abuse.
Violence Violence describes language related to physical actions intended to hurt, injure, damage, or kill someone or something; describes weapons, guns and related entities, such as manufacturers, associations, legislation, and so on.
Self-harm Self-harm describes language related to physical actions intended to purposely hurt, injure, damage one's body or kill oneself.

Classification can be multi-labeled. For example, when a text sample goes through the text moderation model, it could be classified as both Sexual content and Violence.

Severity levels

Every harm category the service applies also comes with a severity level rating. The severity level is meant to indicate the severity of the consequences of showing the flagged content.

Text: The current version of the text model supports the full 0-7 severity scale. By default, the response will output 4 values: 0, 2, 4, and 6. Each two adjacent levels are mapped to a single level. Users could use "outputType" in request and set it as "EightSeverityLevels" to get 8 values in output: 0,1,2,3,4,5,6,7. You can refer text content severity levels definitions for details.

  • [0,1] -> 0
  • [2,3] -> 2
  • [4,5] -> 4
  • [6,7] -> 6

Image: The current version of the image model supports the trimmed version of the full 0-7 severity scale. The classifier only returns severities 0, 2, 4, and 6; each two adjacent levels are mapped to a single level. You can refer image content severity levels definitions for details.

  • [0,1] -> 0
  • [2,3] -> 2
  • [4,5] -> 4
  • [6,7] -> 6

Text blocklist management

Following operations are supported to manage your text blocklist:

  • Create or modify a blocklist
  • List all blocklists
  • Get a blocklist by blocklistName
  • Add blocklistItems to a blocklist
  • Remove blocklistItems from a blocklist
  • List all blocklistItems in a blocklist by blocklistName
  • Get a blocklistItem in a blocklist by blocklistItemId and blocklistName
  • Delete a blocklist and all of its blocklistItems

You can set the blocklists you want to use when analyze text, then you can get blocklist match result from returned response.

Examples

The following section provides several code snippets covering some of the most common Content Safety service tasks, including:

Please refer to sample data for the data used here. For more samples, please refer to samples.

Analyze text

Analyze text without blocklists


    import os
    from azure.ai.contentsafety import ContentSafetyClient
    from azure.ai.contentsafety.models import TextCategory
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError
    from azure.ai.contentsafety.models import AnalyzeTextOptions

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Content Safety client
    client = ContentSafetyClient(endpoint, AzureKeyCredential(key))

    # Construct a request
    request = AnalyzeTextOptions(text="You are an idiot")

    # Analyze text
    try:
        response = client.analyze_text(request)
    except HttpResponseError as e:
        print("Analyze text failed.")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

    hate_result = next(item for item in response.categories_analysis if item.category == TextCategory.HATE)
    self_harm_result = next(item for item in response.categories_analysis if item.category == TextCategory.SELF_HARM)
    sexual_result = next(item for item in response.categories_analysis if item.category == TextCategory.SEXUAL)
    violence_result = next(item for item in response.categories_analysis if item.category == TextCategory.VIOLENCE)

    if hate_result:
        print(f"Hate severity: {hate_result.severity}")
    if self_harm_result:
        print(f"SelfHarm severity: {self_harm_result.severity}")
    if sexual_result:
        print(f"Sexual severity: {sexual_result.severity}")
    if violence_result:
        print(f"Violence severity: {violence_result.severity}")

Analyze text with blocklists


    import os
    from azure.ai.contentsafety import ContentSafetyClient
    from azure.core.credentials import AzureKeyCredential
    from azure.ai.contentsafety.models import AnalyzeTextOptions
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Content Safety client
    client = ContentSafetyClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"
    input_text = "I h*te you and I want to k*ll you."

    try:
        # After you edit your blocklist, it usually takes effect in 5 minutes, please wait some time before analyzing with blocklist after editing.
        analysis_result = client.analyze_text(
            AnalyzeTextOptions(text=input_text, blocklist_names=[blocklist_name], halt_on_blocklist_hit=False)
        )
        if analysis_result and analysis_result.blocklists_match:
            print("\nBlocklist match results: ")
            for match_result in analysis_result.blocklists_match:
                print(
                    f"BlocklistName: {match_result.blocklist_name}, BlockItemId: {match_result.blocklist_item_id}, "
                    f"BlockItemText: {match_result.blocklist_item_text}"
                )
    except HttpResponseError as e:
        print("\nAnalyze text failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Analyze image


    import os
    from azure.ai.contentsafety import ContentSafetyClient
    from azure.ai.contentsafety.models import ImageCategory
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError
    from azure.ai.contentsafety.models import AnalyzeImageOptions, ImageData

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]
    image_path = os.path.abspath(os.path.join(os.path.abspath(__file__), "..", "./sample_data/image.jpg"))

    # Create a Content Safety client
    client = ContentSafetyClient(endpoint, AzureKeyCredential(key))

    # Build request
    with open(image_path, "rb") as file:
        request = AnalyzeImageOptions(image=ImageData(content=file.read()))

    # Analyze image
    try:
        response = client.analyze_image(request)
    except HttpResponseError as e:
        print("Analyze image failed.")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

    hate_result = next(item for item in response.categories_analysis if item.category == ImageCategory.HATE)
    self_harm_result = next(item for item in response.categories_analysis if item.category == ImageCategory.SELF_HARM)
    sexual_result = next(item for item in response.categories_analysis if item.category == ImageCategory.SEXUAL)
    violence_result = next(item for item in response.categories_analysis if item.category == ImageCategory.VIOLENCE)

    if hate_result:
        print(f"Hate severity: {hate_result.severity}")
    if self_harm_result:
        print(f"SelfHarm severity: {self_harm_result.severity}")
    if sexual_result:
        print(f"Sexual severity: {sexual_result.severity}")
    if violence_result:
        print(f"Violence severity: {violence_result.severity}")

Manage text blocklist

Create or update text blocklist


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.ai.contentsafety.models import TextBlocklist
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"
    blocklist_description = "Test blocklist management."

    try:
        blocklist = client.create_or_update_text_blocklist(
            blocklist_name=blocklist_name,
            options=TextBlocklist(blocklist_name=blocklist_name, description=blocklist_description),
        )
        if blocklist:
            print("\nBlocklist created or updated: ")
            print(f"Name: {blocklist.blocklist_name}, Description: {blocklist.description}")
    except HttpResponseError as e:
        print("\nCreate or update text blocklist failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

List text blocklists


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    try:
        blocklists = client.list_text_blocklists()
        if blocklists:
            print("\nList blocklists: ")
            for blocklist in blocklists:
                print(f"Name: {blocklist.blocklist_name}, Description: {blocklist.description}")
    except HttpResponseError as e:
        print("\nList text blocklists failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Get text blocklist


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"

    try:
        blocklist = client.get_text_blocklist(blocklist_name=blocklist_name)
        if blocklist:
            print("\nGet blocklist: ")
            print(f"Name: {blocklist.blocklist_name}, Description: {blocklist.description}")
    except HttpResponseError as e:
        print("\nGet text blocklist failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Delete text blocklist


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"

    try:
        client.delete_text_blocklist(blocklist_name=blocklist_name)
        print(f"\nDeleted blocklist: {blocklist_name}")
    except HttpResponseError as e:
        print("\nDelete blocklist failed:")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Add blockItems


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.ai.contentsafety.models import AddOrUpdateTextBlocklistItemsOptions, TextBlocklistItem
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"
    block_item_text_1 = "k*ll"
    block_item_text_2 = "h*te"

    block_items = [TextBlocklistItem(text=block_item_text_1), TextBlocklistItem(text=block_item_text_2)]
    try:
        result = client.add_or_update_blocklist_items(
            blocklist_name=blocklist_name, options=AddOrUpdateTextBlocklistItemsOptions(blocklist_items=block_items)
        )
        for block_item in result.blocklist_items:
            print(
                f"BlockItemId: {block_item.blocklist_item_id}, Text: {block_item.text}, Description: {block_item.description}"
            )
    except HttpResponseError as e:
        print("\nAdd block items failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

List blockItems


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.core.credentials import AzureKeyCredential
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"

    try:
        block_items = client.list_text_blocklist_items(blocklist_name=blocklist_name)
        if block_items:
            print("\nList block items: ")
            for block_item in block_items:
                print(
                    f"BlockItemId: {block_item.blocklist_item_id}, Text: {block_item.text}, "
                    f"Description: {block_item.description}"
                )
    except HttpResponseError as e:
        print("\nList block items failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Get blockItem


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.core.credentials import AzureKeyCredential
    from azure.ai.contentsafety.models import TextBlocklistItem, AddOrUpdateTextBlocklistItemsOptions
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"
    block_item_text_1 = "k*ll"

    try:
        # Add a blockItem
        add_result = client.add_or_update_blocklist_items(
            blocklist_name=blocklist_name,
            options=AddOrUpdateTextBlocklistItemsOptions(blocklist_items=[TextBlocklistItem(text=block_item_text_1)]),
        )
        if not add_result or not add_result.blocklist_items or len(add_result.blocklist_items) <= 0:
            raise RuntimeError("BlockItem not created.")
        block_item_id = add_result.blocklist_items[0].blocklist_item_id

        # Get this blockItem by blockItemId
        block_item = client.get_text_blocklist_item(blocklist_name=blocklist_name, blocklist_item_id=block_item_id)
        print("\nGet blockitem: ")
        print(
            f"BlockItemId: {block_item.blocklist_item_id}, Text: {block_item.text}, Description: {block_item.description}"
        )
    except HttpResponseError as e:
        print("\nGet block item failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Remove blockItems


    import os
    from azure.ai.contentsafety import BlocklistClient
    from azure.core.credentials import AzureKeyCredential
    from azure.ai.contentsafety.models import (
        TextBlocklistItem,
        AddOrUpdateTextBlocklistItemsOptions,
        RemoveTextBlocklistItemsOptions,
    )
    from azure.core.exceptions import HttpResponseError

    key = os.environ["CONTENT_SAFETY_KEY"]
    endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

    # Create a Blocklist client
    client = BlocklistClient(endpoint, AzureKeyCredential(key))

    blocklist_name = "TestBlocklist"
    block_item_text_1 = "k*ll"

    try:
        # Add a blockItem
        add_result = client.add_or_update_blocklist_items(
            blocklist_name=blocklist_name,
            options=AddOrUpdateTextBlocklistItemsOptions(blocklist_items=[TextBlocklistItem(text=block_item_text_1)]),
        )
        if not add_result or not add_result.blocklist_items or len(add_result.blocklist_items) <= 0:
            raise RuntimeError("BlockItem not created.")
        block_item_id = add_result.blocklist_items[0].blocklist_item_id

        # Remove this blockItem by blockItemId
        client.remove_blocklist_items(
            blocklist_name=blocklist_name, options=RemoveTextBlocklistItemsOptions(blocklist_item_ids=[block_item_id])
        )
        print(f"\nRemoved blockItem: {add_result.blocklist_items[0].blocklist_item_id}")
    except HttpResponseError as e:
        print("\nRemove block item failed: ")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

Troubleshooting

General

Azure AI Content Safety client library will raise exceptions defined in Azure Core. Error codes are defined as below:

Error Code Possible reasons Suggestions
InvalidRequestBody One or more fields in the request body do not match the API definition. 1. Check the API version you specified in the API call.
2. Check the corresponding API definition for the API version you selected.
InvalidResourceName The resource name you specified in the URL does not meet the requirements, like the blocklist name, blocklist term ID, etc. 1. Check the API version you specified in the API call.
2. Check whether the given name has invalid characters according to the API definition.
ResourceNotFound The resource you specified in the URL may not exist, like the blocklist name. 1. Check the API version you specified in the API call.
2. Double check the existence of the resource specified in the URL.
InternalError Some unexpected situations on the server side have been triggered. 1. You may want to retry a few times after a small period and see it the issue happens again.
2. Contact Azure Support if this issue persists.
ServerBusy The server side cannot process the request temporarily. 1. You may want to retry a few times after a small period and see it the issue happens again.
2.Contact Azure Support if this issue persists.
TooManyRequests The current RPS has exceeded the quota for your current SKU. 1. Check the pricing table to understand the RPS quota.
2.Contact Azure Support if you need more QPS.

Logging

This library uses the standard logging library for logging.

Basic information about HTTP sessions (URLs, headers, etc.) is logged at INFO level.

Detailed DEBUG level logging, including request/response bodies and unredacted headers, can be enabled on the client or per-operation with the logging_enable keyword argument.

See full SDK logging documentation with examples here.

Optional Configuration

Optional keyword arguments can be passed in at the client and per-operation level. The azure-core reference documentation describes available configurations for retries, logging, transport protocols, and more.

Next steps

Additional documentation

For more extensive documentation on Azure Content Safety, see the Azure AI Content Safety on docs.microsoft.com.

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.

When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information, see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.